main_backup.py

import os
import asyncio
import httpx
import json
import re
import uuid
import base64
from io import BytesIO
from fastapi import FastAPI, Request
from fastapi.responses import HTMLResponse, JSONResponse, FileResponse
from fastapi.middleware.cors import CORSMiddleware
from datetime import datetime
from pathlib import Path

# python-docx imports
from docx import Document as DocxDocument
from docx.shared import Inches, Pt, RGBColor
from docx.enum.text import WD_ALIGN_PARAGRAPH
from docx.oxml.ns import qn
from docx.oxml import OxmlElement

app = FastAPI()
app.add_middleware(CORSMiddleware, allow_origins=["*"], allow_methods=["*"], allow_headers=["*"])

GOOGLE_API_KEY     = os.getenv("GOOGLE_API_KEY", "")
OPENROUTER_API_KEY = os.getenv("OPENROUTER_API_KEY", "")
GROQ_API_KEY       = os.getenv("GROQ_API_KEY", "")
PEXELS_API_KEY     = os.getenv("PEXELS_API_KEY", "")    # free at pexels.com/api
HF_API_KEY         = os.getenv("HF_TOKEN", "")           # HuggingFace token for image gen

DOCS_DIR = Path("docs")
DOCS_DIR.mkdir(exist_ok=True)

PROVIDERS = {
    "gemini": {
        "name": "Google Gemini", "type": "gemini", "key": GOOGLE_API_KEY,
    },
    "openrouter": {
        "name": "OpenRouter", "type": "openai_compat", "key": OPENROUTER_API_KEY,
        "base_url": "https://openrouter.ai/api/v1/chat/completions",
        "headers": {
            "HTTP-Referer": "https://huggingface.co/spaces/vfven/mission-control-ui",
            "X-Title": "Mission Control AI",
        },
    },
    "groq": {
        "name": "Groq", "type": "openai_compat", "key": GROQ_API_KEY,
        "base_url": "https://api.groq.com/openai/v1/chat/completions",
        "headers": {},
    },
}

# ── DEFAULT AGENTS (can be extended via UI) ───────────────────────────────
DEFAULT_AGENTS = [
    {
        "key": "manager", "name": "Manager", "provider": "gemini",
        "role": "Gerente de proyecto experto en coordinar equipos y planificar estrategias. Cuando el usuario pide un documento o informe, DEBES indicar en tu respuesta qué agentes necesitas activar usando el formato JSON: {\"delegate\": [\"writer\", \"analyst\"]} al final de tu respuesta.",
        "models": ["gemini-2.5-flash-preview-04-17", "gemini-2.0-flash", "gemini-1.5-flash"],
    },
    {
        "key": "developer", "name": "Developer", "provider": "openrouter",
        "role": "Programador senior especialista en crear aplicaciones y soluciones técnicas.",
        "models": ["qwen/qwen3-4b:free", "meta-llama/llama-3.3-70b-instruct:free", "mistralai/mistral-small-3.1-24b-instruct:free", "google/gemma-3-12b-it:free"],
    },
    {
        "key": "analyst", "name": "Analyst", "provider": "openrouter",
        "role": "Analista de negocios experto en evaluar viabilidad, riesgos y oportunidades. También revisa y critica documentos formales.",
        "models": ["meta-llama/llama-3.3-70b-instruct:free", "mistralai/mistral-small-3.1-24b-instruct:free", "google/gemma-3-27b-it:free", "qwen/qwen3-4b:free"],
    },
    {
        "key": "writer", "name": "Writer", "provider": "openrouter",
        "role": "Especialista en redacción de documentos formales, informes ejecutivos y reportes técnicos. Escribe en formato estructurado con secciones claras usando ### para títulos y ** para subtítulos.",
        "models": ["meta-llama/llama-3.3-70b-instruct:free", "mistralai/mistral-small-3.1-24b-instruct:free", "qwen/qwen3-4b:free", "google/gemma-3-12b-it:free"],
    },
    {
        "key": "image_agent", "name": "ImageAgent", "provider": "gemini",
        "role": "Agente especializado en buscar y proveer imágenes relevantes. Cuando se te pida imágenes sobre un tema, responde con una lista JSON de términos de búsqueda en inglés: {\"image_queries\": [\"term1\", \"term2\", \"term3\"]}",
        "models": ["gemini-2.0-flash", "gemini-1.5-flash"],
    },
]

SUBSTITUTE_MODELS = {
    "groq": ["llama-3.3-70b-versatile", "llama3-70b-8192", "gemma2-9b-it"],
}

# Runtime agent registry (can be extended)
agent_registry = {a["key"]: dict(a) for a in DEFAULT_AGENTS}
mission_history = []


# ── LLM CALLERS ───────────────────────────────────────────────────────────
async def call_gemini(model: str, system: str, user: str, key: str) -> str:
    url = f"https://generativelanguage.googleapis.com/v1beta/models/{model}:generateContent?key={key}"
    payload = {
        "contents": [{"role": "user", "parts": [{"text": f"{system}\n\n{user}"}]}],
        "generationConfig": {"maxOutputTokens": 2048, "temperature": 0.7},
    }
    async with httpx.AsyncClient(timeout=90) as client:
        r = await client.post(url, json=payload)
        r.raise_for_status()
        return r.json()["candidates"][0]["content"]["parts"][0]["text"]


async def call_openai_compat(base_url: str, model: str, system: str, user: str,
                              key: str, extra_headers: dict) -> str:
    headers = {"Authorization": f"Bearer {key}", "Content-Type": "application/json", **extra_headers}
    payload = {
        "model": model,
        "messages": [{"role": "system", "content": system}, {"role": "user", "content": user}],
        "max_tokens": 2048, "temperature": 0.7,
    }
    async with httpx.AsyncClient(timeout=90) as client:
        r = await client.post(base_url, json=payload, headers=headers)
        r.raise_for_status()
        return r.json()["choices"][0]["message"]["content"]


async def call_agent_llm(agent: dict, task: str) -> str:
    prov_key = agent["provider"]
    provider = PROVIDERS[prov_key]
    system   = f"Eres {agent['name']}. {agent['role']} Responde en español. Sé conciso y profesional."
    last_err = None

    for model in agent["models"]:
        try:
            if provider["type"] == "gemini":
                return await call_gemini(model, system, task, provider["key"])
            else:
                return await call_openai_compat(
                    provider["base_url"], model, system, task,
                    provider["key"], provider.get("headers", {}))
        except Exception as e:
            last_err = str(e)

    # Groq fallback
    if GROQ_API_KEY:
        groq = PROVIDERS["groq"]
        for m in SUBSTITUTE_MODELS["groq"]:
            try:
                return await call_openai_compat(
                    groq["base_url"], m, system, task, groq["key"], {})
            except Exception as e:
                last_err = str(e)

    raise Exception(f"All providers failed: {last_err}")


# ── IMAGE FETCHING ─────────────────────────────────────────────────────────
async def fetch_pexels_image(query: str) -> bytes | None:
    if not PEXELS_API_KEY:
        return None
    try:
        async with httpx.AsyncClient(timeout=20) as client:
            r = await client.get(
                "https://api.pexels.com/v1/search",
                params={"query": query, "per_page": 1, "orientation": "landscape"},
                headers={"Authorization": PEXELS_API_KEY}
            )
            data = r.json()
            if data.get("photos"):
                img_url = data["photos"][0]["src"]["medium"]
                img_r = await client.get(img_url)
                return img_r.content
    except Exception:
        pass
    return None


async def generate_hf_image(prompt: str) -> bytes | None:
    if not HF_API_KEY:
        return None
    try:
        async with httpx.AsyncClient(timeout=60) as client:
            r = await client.post(
                "https://api-inference.huggingface.co/models/stabilityai/stable-diffusion-xl-base-1.0",
                headers={"Authorization": f"Bearer {HF_API_KEY}"},
                json={"inputs": prompt, "parameters": {"width": 512, "height": 384}},
            )
            if r.status_code == 200 and r.headers.get("content-type", "").startswith("image"):
                return r.content
    except Exception:
        pass
    return None


async def get_image_for_query(query: str) -> bytes | None:
    img = await fetch_pexels_image(query)
    if img:
        return img
    return await generate_hf_image(query)


# ── DOCX BUILDER ──────────────────────────────────────────────────────────
def build_docx(title: str, sections: dict, images: list[bytes], analyst_review: str) -> bytes:
    doc = DocxDocument()

    # Page margins
    for section in doc.sections:
        section.top_margin    = Inches(1)
        section.bottom_margin = Inches(1)
        section.left_margin   = Inches(1.2)
        section.right_margin  = Inches(1.2)

    # Title
    title_para = doc.add_heading(title, 0)
    title_para.alignment = WD_ALIGN_PARAGRAPH.CENTER
    run = title_para.runs[0]
    run.font.color.rgb = RGBColor(0x1a, 0x56, 0xdb)

    # Date + subtitle
    sub = doc.add_paragraph()
    sub.alignment = WD_ALIGN_PARAGRAPH.CENTER
    sub.add_run(f"Mission Control AI — {datetime.now().strftime('%B %d, %Y')}").italic = True
    doc.add_paragraph()

    # Horizontal rule
    p = doc.add_paragraph()
    pPr = p._p.get_or_add_pPr()
    pBdr = OxmlElement("w:pBdr")
    bottom = OxmlElement("w:bottom")
    bottom.set(qn("w:val"), "single")
    bottom.set(qn("w:sz"), "6")
    bottom.set(qn("w:color"), "1a56db")
    pBdr.append(bottom)
    pPr.append(pBdr)

    img_index = 0

    # Writer content sections
    writer_text = sections.get("writer", "")
    current_lines = []

    def flush_lines():
        nonlocal current_lines
        if current_lines:
            para_text = " ".join(current_lines).strip()
            if para_text:
                p = doc.add_paragraph(para_text)
                p.paragraph_format.space_after = Pt(6)
            current_lines = []

    for line in writer_text.split("\n"):
        stripped = line.strip()
        if not stripped:
            flush_lines()
            continue
        if stripped.startswith("### "):
            flush_lines()
            h = doc.add_heading(stripped[4:], level=2)
            # Insert image after each major section heading if available
            if img_index < len(images) and images[img_index]:
                try:
                    img_stream = BytesIO(images[img_index])
                    doc.add_picture(img_stream, width=Inches(5))
                    last_para = doc.paragraphs[-1]
                    last_para.alignment = WD_ALIGN_PARAGRAPH.CENTER
                    cap = doc.add_paragraph(f"Figure {img_index + 1}")
                    cap.alignment = WD_ALIGN_PARAGRAPH.CENTER
                    cap.runs[0].italic = True
                    cap.runs[0].font.size = Pt(9)
                    img_index += 1
                except Exception:
                    pass
        elif stripped.startswith("## "):
            flush_lines()
            doc.add_heading(stripped[3:], level=1)
        elif stripped.startswith("**") and stripped.endswith("**"):
            flush_lines()
            p = doc.add_paragraph()
            run = p.add_run(stripped.strip("**"))
            run.bold = True
        elif stripped.startswith("- ") or stripped.startswith("* "):
            flush_lines()
            doc.add_paragraph(stripped[2:], style="List Bullet")
        else:
            current_lines.append(stripped)

    flush_lines()

    # Remaining images
    while img_index < len(images):
        if images[img_index]:
            try:
                img_stream = BytesIO(images[img_index])
                doc.add_picture(img_stream, width=Inches(5))
                last_para = doc.paragraphs[-1]
                last_para.alignment = WD_ALIGN_PARAGRAPH.CENTER
                img_index += 1
            except Exception:
                img_index += 1
        else:
            img_index += 1

    # Analyst review section
    if analyst_review:
        doc.add_page_break()
        doc.add_heading("Análisis y Revisión", level=1)
        for line in analyst_review.split("\n"):
            if line.strip():
                doc.add_paragraph(line.strip())

    # Footer
    doc.add_paragraph()
    footer_p = doc.add_paragraph()
    footer_p.alignment = WD_ALIGN_PARAGRAPH.CENTER
    footer_run = footer_p.add_run("— Generado por Mission Control AI —")
    footer_run.italic = True
    footer_run.font.size = Pt(9)
    footer_run.font.color.rgb = RGBColor(0x6b, 0x72, 0x80)

    buf = BytesIO()
    doc.save(buf)
    buf.seek(0)
    return buf.read()


# ── PARSE MANAGER DELEGATION ───────────────────────────────────────────────
def parse_delegation(manager_text: str) -> list[str]:
    match = re.search(r'\{[^}]*"delegate"\s*:\s*\[([^\]]*)\][^}]*\}', manager_text)
    if match:
        raw = match.group(1)
        keys = re.findall(r'"(\w+)"', raw)
        return keys
    # Heuristic fallback: detect keywords in manager response
    delegates = []
    lower = manager_text.lower()
    if any(w in lower for w in ["informe", "documento", "redact", "escrib", "report", "word"]):
        delegates.extend(["writer"])
    if any(w in lower for w in ["imagen", "image", "foto", "visual", "ilustr"]):
        delegates.extend(["image_agent"])
    if any(w in lower for w in ["analiz", "revisar", "evalua", "critic"]):
        delegates.extend(["analyst"])
    return list(dict.fromkeys(delegates))  # deduplicate


def parse_image_queries(image_agent_text: str) -> list[str]:
    match = re.search(r'"image_queries"\s*:\s*\[([^\]]*)\]', image_agent_text)
    if match:
        return re.findall(r'"([^"]+)"', match.group(1))
    return []


def clean_text(text: str) -> str:
    # Remove JSON blocks from display text
    text = re.sub(r'\{[^}]*"delegate"[^}]*\}', '', text)
    text = re.sub(r'\{[^}]*"image_queries"[^}]*\}', '', text)
    return text.strip()


# ── ROUTES ─────────────────────────────────────────────────────────────────
@app.get("/", response_class=HTMLResponse)
async def root():
    return HTMLResponse(Path("templates/index.html").read_text())


@app.get("/api/agents")
async def get_agents():
    return {"agents": [
        {"key": a["key"], "name": a["name"], "role": a["role"]}
        for a in agent_registry.values()
    ]}


@app.post("/api/agents/add")
async def add_agent(request: Request):
    body = await request.json()
    key  = re.sub(r'\W+', '_', body.get("key", "").lower().strip())
    if not key:
        return JSONResponse({"error": "key required"}, status_code=400)
    agent_registry[key] = {
        "key":      key,
        "name":     body.get("name", key.capitalize()),
        "role":     body.get("role", "Agente de propósito general."),
        "provider": body.get("provider", "openrouter"),
        "models":   body.get("models", ["meta-llama/llama-3.3-70b-instruct:free"]),
    }
    return {"success": True, "agent": agent_registry[key]}


@app.delete("/api/agents/{key}")
async def delete_agent(key: str):
    if key in ("manager", "developer", "analyst"):
        return JSONResponse({"error": "Cannot delete core agents"}, status_code=400)
    if key in agent_registry:
        del agent_registry[key]
    return {"success": True}


@app.get("/api/history")
async def get_history():
    return {"history": mission_history[-20:]}


@app.get("/api/docs/{filename}")
async def download_doc(filename: str):
    path = DOCS_DIR / filename
    if not path.exists() or not filename.endswith(".docx"):
        return JSONResponse({"error": "File not found"}, status_code=404)
    return FileResponse(
        path,
        media_type="application/vnd.openxmlformats-officedocument.wordprocessingml.document",
        filename=filename,
    )


@app.post("/api/mission")
async def run_mission(request: Request):
    body = await request.json()
    task = body.get("task", "").strip()
    if not task:
        return JSONResponse({"error": "No task provided"}, status_code=400)

    started_at = datetime.now().isoformat()
    results    = {}
    doc_file   = None
    events     = []   # orchestration event log

    def log(msg: str):
        events.append({"time": datetime.now().strftime("%H:%M:%S"), "msg": msg})

    # ── STEP 1: Manager plans ──────────────────────────────────────────────
    log("Manager analyzing task...")
    manager_agent = agent_registry["manager"]
    try:
        manager_raw = await call_agent_llm(manager_agent, task)
        delegates   = parse_delegation(manager_raw)
        manager_msg = clean_text(manager_raw)
        results["manager"] = {
            "status": "active", "message": manager_msg,
            "model": manager_agent["models"][0],
            "delegates": delegates,
        }
        log(f"Manager delegated to: {delegates or 'none'}")
    except Exception as e:
        results["manager"] = {"status": "resting", "message": str(e), "model": ""}
        delegates = []
        log(f"Manager failed: {e}")

    # ── STEP 2: Run delegated agents ───────────────────────────────────────
    image_bytes  = []
    writer_text  = ""
    analyst_text = ""

    async def run_delegated(key: str):
        nonlocal writer_text, analyst_text
        if key not in agent_registry:
            log(f"Agent '{key}' not found in registry")
            return

        agent = agent_registry[key]
        log(f"{agent['name']} starting...")

        if key == "image_agent":
            try:
                img_prompt = f"Proporciona consultas de búsqueda de imágenes en inglés para ilustrar: {task}"
                raw = await call_agent_llm(agent, img_prompt)
                queries = parse_image_queries(raw)
                if not queries:
                    queries = [task[:40]]
                log(f"ImageAgent searching: {queries}")
                imgs = await asyncio.gather(*[get_image_for_query(q) for q in queries[:3]])
                for img in imgs:
                    if img:
                        image_bytes.append(img)
                results["image_agent"] = {
                    "status": "active",
                    "message": f"Encontradas {len(image_bytes)} imágenes para: {', '.join(queries[:3])}",
                    "model": agent["models"][0],
                }
                log(f"ImageAgent: {len(image_bytes)} images found")
            except Exception as e:
                results["image_agent"] = {"status": "resting", "message": str(e), "model": ""}
                log(f"ImageAgent failed: {e}")

        elif key == "writer":
            try:
                writer_prompt = (
                    f"Redacta un informe formal y completo sobre: {task}\n\n"
                    "Usa este formato:\n"
                    "## Resumen Ejecutivo\n[contenido]\n\n"
                    "### Introducción\n[contenido]\n\n"
                    "### Desarrollo\n[contenido con subsecciones]\n\n"
                    "### Conclusiones\n[contenido]\n\n"
                    "### Recomendaciones\n[contenido]\n\n"
                    "Sé formal, detallado y profesional."
                )
                writer_text = await call_agent_llm(agent, writer_prompt)
                results["writer"] = {
                    "status": "active",
                    "message": writer_text[:200] + "..." if len(writer_text) > 200 else writer_text,
                    "model": agent["models"][0],
                }
                log("Writer completed document")
            except Exception as e:
                results["writer"] = {"status": "resting", "message": str(e), "model": ""}
                log(f"Writer failed: {e}")

        elif key == "analyst":
            try:
                content_to_review = writer_text or manager_msg
                analyst_prompt = (
                    f"Revisa y analiza el siguiente contenido sobre: {task}\n\n"
                    f"Contenido:\n{content_to_review[:1000]}\n\n"
                    "Proporciona: 1) Evaluación de calidad, 2) Puntos fuertes, "
                    "3) Áreas de mejora, 4) Conclusión final."
                )
                analyst_text = await call_agent_llm(agent, analyst_prompt)
                results["analyst"] = {
                    "status": "active",
                    "message": analyst_text[:200] + "..." if len(analyst_text) > 200 else analyst_text,
                    "model": agent["models"][0],
                }
                log("Analyst review completed")
            except Exception as e:
                results["analyst"] = {"status": "resting", "message": str(e), "model": ""}
                log(f"Analyst failed: {e}")

        else:
            # Generic agent
            try:
                raw = await call_agent_llm(agent, task)
                results[key] = {"status": "active", "message": raw, "model": agent["models"][0]}
                log(f"{agent['name']} completed")
            except Exception as e:
                results[key] = {"status": "resting", "message": str(e), "model": ""}

    # Run image_agent and writer in parallel, then analyst
    parallel_first = [k for k in delegates if k in ("writer", "image_agent")]
    sequential_after = [k for k in delegates if k == "analyst"]
    other = [k for k in delegates if k not in ("writer", "image_agent", "analyst")]

    if parallel_first or other:
        await asyncio.gather(*[run_delegated(k) for k in parallel_first + other])

    if sequential_after:
        for k in sequential_after:
            await run_delegated(k)

    # ── STEP 3: Build .docx if writer was involved ─────────────────────────
    if writer_text:
        log("Assembling Word document...")
        try:
            doc_bytes = build_docx(task, {"writer": writer_text}, image_bytes, analyst_text)
            safe_name = re.sub(r'[^\w\-]', '_', task[:40])
            doc_filename = f"{safe_name}_{datetime.now().strftime('%Y%m%d_%H%M%S')}.docx"
            (DOCS_DIR / doc_filename).write_bytes(doc_bytes)
            doc_file = doc_filename
            results["manager"]["doc_file"] = doc_filename
            log(f"Document ready: {doc_filename}")
        except Exception as e:
            log(f"Document build failed: {e}")

    # Mark idle agents
    for key in agent_registry:
        if key not in results:
            results[key] = {"status": "idle", "message": "", "model": ""}

    final = results.get("manager", {}).get("message", "")[:300]

    entry = {
        "id": len(mission_history) + 1,
        "task": task,
        "started_at": started_at,
        "ended_at": datetime.now().isoformat(),
        "results": results,
        "final": final,
        "doc_file": doc_file,
        "events": events,
    }
    mission_history.append(entry)

    return JSONResponse({
        "success": True, "task": task,
        "results": results, "final": final,
        "doc_file": doc_file, "events": events,
        "mission_id": entry["id"],
    })


@app.get("/api/health")
async def health():
    return {
        "status": "ok",
        "providers": {
            "gemini":     "ok" if GOOGLE_API_KEY     else "missing",
            "openrouter": "ok" if OPENROUTER_API_KEY else "missing",
            "groq":       "ok" if GROQ_API_KEY       else "missing",
            "pexels":     "ok" if PEXELS_API_KEY     else "missing (optional)",
            "hf_images":  "ok" if HF_API_KEY         else "missing (optional)",
        }
    }